large scale Baird/Nipkow Televisor project "Big Paul"

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Steve Anderson, Dave Moll, Andrew Davie

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Wed Mar 07, 2018 12:55 am

Andrew Davie wrote:I think it's mesmerising, awesome and incredibly dangerous in equal measures.
Brightness looks good!


Thanks! Brightness should get better, once we found a driver that can power the LED at its nominal intensity of 25W. Yes, very dangerous indeed, that's why I only dare to film off the control monitor, and only dare to peek around the corner with the camera to get an actual glimpse of the spinning disk. But the cage will be massive, here are some photos:

Mqhf5d00SqeP9KUOmzjErQ.jpg


J3%2uV1tR0eYP%HYNCHT4g.jpg


vVao6XcWT3Gy%RK17rtoqQ.jpg


(I am not sure why the orientation is wrong in the previews)
Last edited by gebseng on Wed Mar 07, 2018 10:18 am, edited 1 time in total.
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Robonz » Wed Mar 07, 2018 10:02 am

Hi Gebhard

Nice progress. As far as driving the light source...

I guess you need to spec the maximum frequency response you expect the light source to do. The current the switch/transistor needs to manage. Analog or digital light control. If you go the digital path which will run much cooler, here is my favourite chip.

9 amp fet drivers - TC4421 (thats 9 amps in the gate of the fet, the output can be much larger)

http://ww1.microchip.com/downloads/en/DeviceDoc/21946B.pdf

Put around 2 ohms of gate resistor and use any large/fast mosfet. That's not a big deal as far as current goes. Your main limit is how fast you want to switch the led on and off. Smaller gate capacitance is key if you want fast switching. The fet driver makes the fast switching a whole bunch easier. I have used these and found them really nice to work with.

This idea requires PWM so you would need an analog to PWM module if you are starting with analog. Linear is an option too but things get much hotter and harder to manage.


Cheers
Keith
User avatar
Robonz
Evil Genius
 
Posts: 147
Joined: Sat Aug 12, 2017 7:15 pm
Location: New Zealand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Wed Mar 07, 2018 10:26 am

Robonz wrote:Hi Gebhard

Nice progress. As far as driving the light source...

I guess you need to spec the maximum frequency response you expect the light source to do. The current the switch/transistor needs to manage. Analog or digital light control. If you go the digital path which will run much cooler, here is my favourite chip.

9 amp fet drivers - TC4421 (thats 9 amps in the gate of the fet, the output can be much larger)

http://ww1.microchip.com/downloads/en/DeviceDoc/21946B.pdf

Put around 2 ohms of gate resistor and use any large/fast mosfet. That's not a big deal as far as current goes. Your main limit is how fast you want to switch the led on and off. Smaller gate capacitance is key if you want fast switching. The fet driver makes the fast switching a whole bunch easier. I have used these and found them really nice to work with.

This idea requires PWM so you would need an analog to PWM module if you are starting with analog. Linear is an option too but things get much hotter and harder to manage.


Cheers
Keith


Thanks Keith! Do you think that a PWM based system would be fast enough for a potential resolution of 240 lines? I am really interested to hear more about the circuit, if you can point me in the right direction.

We thought about using this LED:
https://www.digikey.com/product-detail/ ... ND/6012737
because it is bright and has a good lumen/watt ratio.

And this photo diode:
https://www.digikey.com/products/en?key ... -22-021-ND
because it is very light sensitive in the visible spectrum, has a very low noise floor and switches fast

Does that make any sense?

thanks,

Gebhard
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Robonz » Wed Mar 07, 2018 11:16 am

You really need to do some calculations to see what frequency response you need. I could probably help there if needed.

Basically you need to calculate how many lines are drawn per second and then we can work out how much detail each line should hold.

So its 240 lines, what is the frame rate? 600 rpm = 10 revolutions per second. Is that just one spiral? I need to work it through when I am not in a rush.

I have not read the thread in detail. What is the photo diode for? Is that for your speed/sync sensor?
User avatar
Robonz
Evil Genius
 
Posts: 147
Joined: Sat Aug 12, 2017 7:15 pm
Location: New Zealand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Wed Mar 07, 2018 5:25 pm

Robonz wrote:You really need to do some calculations to see what frequency response you need. I could probably help there if needed.

Basically you need to calculate how many lines are drawn per second and then we can work out how much detail each line should hold.

So its 240 lines, what is the frame rate? 600 rpm = 10 revolutions per second. Is that just one spiral? I need to work it through when I am not in a rush.

I have not read the thread in detail. What is the photo diode for? Is that for your speed/sync sensor?


Thanks again Keith, I really appreciate your help! The maximum would be 240vertical lines=holes, a horizontal (per line) resolution of 500 dots, and 15frames/second (900rpm). Would that be 1.8 GHz? I am not sure how to factor in the brightness levels per dot.

The photo diode is the imaging element for the camera side, I built two of those televisors.
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Steve Anderson » Wed Mar 07, 2018 5:44 pm

A quick few stabs at the calculator...240 lines at 15fps=3600 lines per second. That multiplied by 500 pixels=a pixel rate of 1.8 million per second. Divided by 2 (a pixel is one half of a sine-wave...sort of) = bandwidth of 900kHz. Add a few percent for blanking/syncs and we'll call it quits at 1MHz. Not too onerous, and certainly easier than 1.8GHz! Brightness doesn't factor in this at all. Kell factor might. But the good news is allowing for (say) a Kell factor of 0.7 reduces the bandwidth to around 700kHz.

But, the above is also altered by the aspect ratio of the active picture area. Kell factor is explained here...

https://en.wikipedia.org/wiki/Kell_factor

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 3843
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Robonz » Wed Mar 07, 2018 8:07 pm

Good stuff Steve. Well that rules out PWM. At 128 levels of video levels that would equate to 1Mhz x 128 = 128Mhz PWM frequency. Note really practical. You are probably best to go old school and make a linear amplifier that can do 1 to 2 Mhz, not something I have done before. I guess it would need to be push pull because I suspect the capacitance of the LED will come into play and the LED will need to be forcefully switched off.

I would be interested to see what a 1Mhz amplifier capable of 25 watts would look like, maybe Steve has some ideas. I guess it would be your typical class A/B amplifier but special due to the frequency. A good challenge for sure. Class A would probably be simpler but you would need to get rid of 25 watts of heat. Looks like you have plenty of places to get some cooling with all that metal.

As far as the photo-diode goes I suspect the classic instrumentation amplifier would be a good place to start. It will need plenty of knobs for black level and brightness as a minimum. I have used this amplifier for fast photo diodes. You do need really smooth power supplies due to the gain needed. You need shielding/short cables on the photodiode.

When choosing the photo-diode, its all about the optics. e.g. are you planning to put lenses in the Nipkow? I suspect you will need some good optical gain otherwise the sensitivity will be extremely low. Trying to turn fractions of light into a signal is tricky. Its very hard to beat the performance of the old classic BPW34. It has a very large sensor area. If you can get the light on to it, it has very good performance.
https://www.vishay.com/docs/81521/bpw34.pdf

instr-amp.gif
instr-amp.gif (8.07 KiB) Viewed 5626 times
User avatar
Robonz
Evil Genius
 
Posts: 147
Joined: Sat Aug 12, 2017 7:15 pm
Location: New Zealand

Linear amplifier 1Mhz

Postby Robonz » Wed Mar 07, 2018 8:52 pm

I couldn't help myself and had to look at this. The LT1210 looks interesting. You could probably put two or three in parallel with an output resistor on each chip. They will drive a 10 ohm load at 15 volts.

15 volts into 10 ohms is ~22 watts. De-rate for cooling, say 12 watts. Two of these would do around 24 watts. They look to have good gain at 1Mhz. They seemed to be the sweet spot between getting really expensive and performance.

LT1210
http://cds.linear.com/docs/en/datasheet/1210fb.pdf

Is it a good idea Steve or did I miss something?
User avatar
Robonz
Evil Genius
 
Posts: 147
Joined: Sat Aug 12, 2017 7:15 pm
Location: New Zealand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Thu Mar 08, 2018 1:48 am

Thanks Keith and Steve!

I compared the specs of your Photo diode suggestion to mine, it looks to be pretty similar. Thanks for the suggestion, I will definitively also order this one.

I think that using the full 25W of my LED is not the main objective at the moment, since the image is already pretty bright at the current 5W, even in a room that is not darkened. If I can get it to 10W that should be plenty for the moment.

As for the camera side:
I was hoping not having to use lenses in the disk, and just find a circuit that is sensitive enough to work with my 0.5mm holes. Does that seem unrealistic?
My idea was to try the flying spot method, which allows me to place a very bright, unmodulated light source behind the disk, and get a light pattern that scans the viewers face in front of the disk. Did I get that right?
Is it correct that I in this case I will need a point-like light source (with a lit area as small as possible), or can I place a condenser (like the one in a slide projector) between light source and disk?

best,

Gebhard
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Steve Anderson » Thu Mar 08, 2018 12:46 pm

Robonz wrote:Well that rules out PWM. At 128 levels of video levels that would equate to 1Mhz x 128 = 128Mhz PWM frequency.

I'm not quite sure where the working out above comes from. I would have thought that (say) 5MHz of PWM would be adequate. Steve Ostler used PWM at 100kHz in his colour mechanical monitor, the bandwidth of the signal was around 10-15kHz as a guess.

In this case the signal was analogue, not derived from a D-A which (in theory) means there's an infinite number of levels between black and white for an analogue signal.

However, even a 5MHz PWM circuit is not for the faint-hearted! And yes, the LED junction capacitance does rear its ugly head. The same applies to photodiodes, but here it's easier. Just shove the output of the photodiode into a zero-impedance load - a transimpedance amplifier, e.g. a grounded-base transistor being the simplest.

The BPW34 is a good choice, I have used it and others have too. But make sure you get the transparent version with the clear case, there's also an IR version which is black. The pdf for the IR version is attached.

Steve A.
Attachments
BPW34 FA.pdf
(63.67 KiB) Downloaded 190 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 3843
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Robonz » Thu Mar 08, 2018 7:50 pm

Hi Steve

Yes good point. The PWM frequency does not need to be that high. But seeing as the source will be analog it would require an analog to PWM converter and cleanly switching a power mosfet at a minimum of 5Mhz would require extra hoops to be jumped through. The LT1210 would be simple in comparison. Just a simple power op amp configuration.http://cds.linear.com/docs/en/datasheet/1210fb.pdf

Yes the BPW34 needs to be the clear case version which is the one I linked to. An IR picture would not look good haha. Especially with a white light source.

Gebhard, a flying spot is an interesting approach and fits the period well. The images I have seen always look a bit odd I guess due to the light always aiming directly at the target. But it is a simple way to get started. The BPW34 will definitely work well for a flying spot as it will pick up much more light at a much cheaper price than the sensor you were looking at. I think you will need a very bright well focused light source behind the Nipkow disc. The BPW34 may need a wide lens and a good light shroud too.

I tried putting a "focused" light source behind my Nipkow disc and can see a clear scanning spot. You need the apertures as thin as possible as parallax will obscure the the light especially on the outer lines. You will find the apertures will need to be bigger than you might think. If you stick a target screen up at your ideal focus you can pencil in the light spots to see the coverage. My tests have shown that laser cut 0.4mm black polypropylene sheet shows great promise for ideal apertures, cheap and simple. I found that putting wet newspaper on both sides before cutting makes clean non melted holes as the wet newspaper quenches the cut. I need to finish my new disc and report on the performance.

Its really great to see this project progressing.

Cheers
Keith
User avatar
Robonz
Evil Genius
 
Posts: 147
Joined: Sat Aug 12, 2017 7:15 pm
Location: New Zealand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Steve Anderson » Thu Mar 08, 2018 8:34 pm

You could drop the PWM frequency to 2MHz (Shannon-Nyquist theorem), but even so, that's still quite a challenge. It can be done but probably with discrete components rather than chips, unless exotic (read expensive). Linear is probably the way to go here, even with the inefficiencies involved. A large part hangs on the LEDs proposed and the quantity of them.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 3843
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Fri Mar 09, 2018 6:13 am

Steve Anderson wrote:You could drop the PWM frequency to 2MHz (Shannon-Nyquist theorem), but even so, that's still quite a challenge. It can be done but probably with discrete components rather than chips, unless exotic (read expensive). Linear is probably the way to go here, even with the inefficiencies involved. A large part hangs on the LEDs proposed and the quantity of them.

Steve A.


Is there a chance that you create a kind of interference between the PWM frequency and the Nipkow disk, if that frequency is too low?
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby gebseng » Fri Mar 09, 2018 6:18 am

Robonz wrote:Hi Steve

Yes good point. The PWM frequency does not need to be that high. But seeing as the source will be analog it would require an analog to PWM converter and cleanly switching a power mosfet at a minimum of 5Mhz would require extra hoops to be jumped through. The LT1210 would be simple in comparison. Just a simple power op amp configuration.http://cds.linear.com/docs/en/datasheet/1210fb.pdf

Yes the BPW34 needs to be the clear case version which is the one I linked to. An IR picture would not look good haha. Especially with a white light source.

Gebhard, a flying spot is an interesting approach and fits the period well. The images I have seen always look a bit odd I guess due to the light always aiming directly at the target. But it is a simple way to get started. The BPW34 will definitely work well for a flying spot as it will pick up much more light at a much cheaper price than the sensor you were looking at. I think you will need a very bright well focused light source behind the Nipkow disc. The BPW34 may need a wide lens and a good light shroud too.

I tried putting a "focused" light source behind my Nipkow disc and can see a clear scanning spot. You need the apertures as thin as possible as parallax will obscure the the light especially on the outer lines. You will find the apertures will need to be bigger than you might think. If you stick a target screen up at your ideal focus you can pencil in the light spots to see the coverage. My tests have shown that laser cut 0.4mm black polypropylene sheet shows great promise for ideal apertures, cheap and simple. I found that putting wet newspaper on both sides before cutting makes clean non melted holes as the wet newspaper quenches the cut. I need to finish my new disc and report on the performance.

Its really great to see this project progressing.

Cheers
Keith


Thanks for the op amp and photo diode tips!

I also tried to just aim my iphone flashlight behind the spinning disk, and got a sharp looking projected pattern.

I am not clear on what you mean by "aperture", is there a second, slit-like aperture you need in front of the nipkow disk? Or do you mean just the holes in the disk?

best,

geb
User avatar
gebseng
Laboratory Assistant
 
Posts: 33
Joined: Sun Mar 05, 2017 9:45 pm
Location: Vienna, Austria

Re: large scale Baird/Nipkow Televisor project "Big Paul"

Postby Klaas Robers » Fri Mar 09, 2018 8:19 am

For the flying spot source there is no need for a fine point light source. The light source should illuminate the picture area of the Nipkow disc evenly. Look what happens in a slide projector In fact the flying spot source is a slide projector, where the Nipkow disc is the slide. It projects a focussed moving spot on the scene.

The light sensors around the scene have the same behaviour as the lamps on a film or TV scene. You can have spot-lights, more or less focussed, or flood lights. It is only the running direction of the light that is reversed. You need also more light sensors to eliminate sharp "shaddows". The faders in the output signals of the sensors are comparable in function to the dimmers of the studio lighting in a TV studio.
User avatar
Klaas Robers
Frankenstein was my uncle.
 
Posts: 1453
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

PreviousNext

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 2 guests